kafka down controller, didn't recive`finishmark` table

[River2000i]

What did you do?

cdc-kafka-controller-failure

// Create changefeed with `open-protocol`
// Run Kafka Consumer
// Create database in the upstream TiDB Cluster
// Pause the changefeed
// Run workload prepare phase.
// Resume the changefeed
// Parallel: periodically inject pod failure chaos into Kafka controller, and run workload phase `run`
// Create database -> run workload -> send finishmark
// Check consistency between source <-> target

What did you expect to see?

finishmark table exists

What did you see instead?

finishmark table not exists

Versions of the cluster

Upstream TiDB cluster version (execute SELECT tidb_version(); in a MySQL client):

(paste TiDB cluster version here)

Upstream TiKV version (execute tikv-server --version):

(paste TiKV version here)

TiCDC version (execute cdc version):

(paste TiCDC version here)